This Week in AI

Microsoft, SentinelOne, and Palo Alto Networks have opted out of MITRE ATT&CK 2025, citing a desire to refocus on innovation; but insiders see deeper issues.

Check Point revealed an attack by Nimbus Manticore, targeting aerospace, defense, and telecom in Denmark, Sweden, and Portugal.

Despite warnings from Google’s John Mueller, some plugins and tools still push LLMs.txt, a non-standard protocol that no major AI system uses.

Yale’s three major AI clubs have seen explosive growth with AIA’s mailing list jumping to 400 students and YAIPI welcoming 40 new fellows.

Security scores for top open-source GenAI models jumped from 1.8% to 99.6% when enhanced with guardrails.

A staggering 86% of healthcare orgs report AI tools deployed without IT oversight.